AI Infrastructure Observatory

Where the AI buildout touches the ground: 152 sites in 45 countries, each with its water, grid-carbon, and community record.

Sites tracked

152

125 operational, 12 under construction and 15 announced

Operational capacity

19.9 GW

sum of reported and estimated site capacity

Building & announced

9.7 GW

full build-out figures for pipeline sites

In water-stressed areas

54

sites in high or extremely high water stress

Capacity figures exist for 123 of 152 sites and mix disclosed, permitted, and estimated values. Each record carries its sources and a confidence rating; see the methodology for how sites are selected and characterized.

Loading map…

The gigawatt era

The ten largest campuses on record. Bars show full build-out capacity, so a mostly-announced campus can dwarf anything running today; color marks how much of it is real yet.

  1. Meta Hyperion (Richland Parish, Louisiana)5.0 GW

    Meta Platforms Inc. · Under construction · Richland Parish, Louisiana, US

  2. AWS Project Rainier (New Carlisle, Indiana)2.2 GW

    Amazon Web Services (Amazon.com Inc.) · Operational · Northern Indiana, US

  3. OpenAI / Oracle Stargate I (Abilene, Texas)1.2 GW

    OpenAI Inc. · Operational · Texas, US

  4. Google / Adani AI Data Center Campus — Visakhapatnam1.0 GW

    Google (Alphabet Inc.) · Announced · Andhra Pradesh, India

  5. xAI Colossus 2 (Southaven, Mississippi)1.0 GW

    xAI · Operational · Southaven, DeSoto County, Mississippi

  6. Amazon Web Services Ashburn (US-EAST-1)0.6 GW

    Amazon Web Services (Amazon.com Inc.) · Operational · Northern Virginia

  7. Amazon Web Services Oregon (Boardman)0.6 GW

    Amazon Web Services (Amazon.com Inc.) · Operational · Pacific Northwest, US

  8. Meta Odense AI Data Center0.6 GW

    Meta Platforms Inc. · Operational · Odense, Funen, Denmark

  9. OpenAI / NEXTDC Sydney S7 AI Supercluster (Eastern Creek)0.6 GW

    OpenAI Inc. · Announced · Sydney, New South Wales, Australia

  10. Meta / Sify Technologies AI Data Center — Visakhapatnam0.5 GW

    Meta Platforms Inc. · Announced · Andhra Pradesh, India

Built where water is scarce

54 of 152 tracked sites — about 6.4 GW of capacity — sit in regions the WRI Aqueduct atlas rates as high or extremely high baseline water stress. Data centers in these places compete with cities and farms for the same water.

Largest eight by capacity shown. Water stress is researcher-assigned per site from WRI Aqueduct 4.0; it describes the basin, not the facility's own draw.

Classification

Explicit AI Infrastructure
Purpose-built for AI training or inference workloads.
AI-Capable Hyperscale
General hyperscale data centers with significant AI workloads.
Associated Infrastructure
Substations, transmission, and power projects serving AI sites.

Environmental data

All environmental values (water stress, grid carbon intensity, drought risk) are researcher-assigned from published sources. Values are never fetched from live APIs. Each value displays its source and confidence tier.

What we do not claim

We do not attribute environmental harm to specific facilities. Community concern flags record events reported in public sources. They do not imply causation. See the methodology for full definitions.

Static snapshot published July 19, 2026. Each site records its own last-verified date; environmental values are researcher-assigned from cited sources, never fetched live.